Properties of Porcine and Recombinant Human Collagen Matrices for

May 10, 2006 - various stoichiometries and collagen concentrations (5-20 w/w %). The resulting hydrogels were clear and, when used as cell growth matr...
0 downloads 0 Views 360KB Size
Biomacromolecules 2006, 7, 1819-1828

1819

Properties of Porcine and Recombinant Human Collagen Matrices for Optically Clear Tissue Engineering Applications Y. Liu,†,‡ M. Griffith,§ M. A. Watsky,| J. V. Forrester,⊥ L. Kuffova´ ,⊥ D. Grant,§ K. Merrett,†,§ and D. J. Carlsson*,†,# National Research Council Canada, 1200 Montreal Road, Ottawa, Canada ON K1A 0R6, University of Ottawa, Eye Institute, 501 Smyth Road, Ottawa, Canada ON K1H 8L6, University of Tennessee Health Center, Memphis, Tennessee, University of Aberdeen, Department of Ophthalmology, Institute of Medical Science, Aberdeen, Scotland AB25 2ZD, and University of Ottawa, Department of Ophthalmology, 451 Smyth Road, Ottawa, Canada K1H 8M5 Received February 20, 2006; Revised Manuscript Received March 23, 2006

Porcine and recombinant human atelocollagen I solutions were cross-linked with a water soluble carbodiimide at various stoichiometries and collagen concentrations (5-20 w/w %). The resulting hydrogels were clear and, when used as cell growth matrices, allowed cell and nerve visualization in vitro and in vivo. We have previously reported that, after six months of implantation in pigs’ and rabbits’ corneas, these robust hydrogels allowed regeneration of host cells and nerves to give optically clear corneas with no detected loss in thickness, indicating stable engraftment. Here, the biocompatible hydrogel formulations leading to this novel in vivo performance were characterized for amine consumption, gel hydration, thermal properties, optical clarity, refractive index, nutrient diffusion, biodegradation, tensile measurements, and average pore diameters. Gels with excellent in vitro (epithelial overgrowth, neurite penetration) and in vivo performance (clarity, touch sensitivity regeneration) had 4-11 nm pores, yet had glucose and albumin diffusive coefficients similar to mammalian corneas and allowed neurite extension through the gels. A variety of synthetic and naturally derived materials have been used to form tissue engineering (TE) scaffolds. Natural soft tissues are hydrogels; thus, synthetic and modified biopolymer hydrogels are an obvious starting point for soft tissue repair or replacement. Collagen-based biomaterials have been widely used in TE.1-10 Collagen I is now readily available, for example, from bovine, porcine, recombinant, or other collagen sources and is an interesting starting material for TE scaffolds. Specifically, collagen contains the cell adhesion peptide motif RGD, and collagen gives robust hydrogels at modest concentrations as a result of its semirigid-rod, triple helix structure. However, an increased turnover of collagen in vivo after wounding or surgery can be expected to result from matrix metalloproteinases (including collagenase enzymes).11 These enzymes are upregulated by loss or damage of the surface epithelium and by inflammation.11 Consequently, any collagen-based TE material should be adequately stabilized to reduce untimely biodegradation. It is well-established that collagen biodegradation can be retarded by chemical cross-linking methods, which also provide materials with enhanced mechanical properties for implantation.2,4,6,7,9 Water-soluble carbodiimides (WSCs) are a family of protein cross-linking reagents which promote the formation of zerolength cross-links by facilitating the aqueous-phase reaction between collagen’s amine and carboxylic acid side-groups to form covalent, amide bonds.1,2,6-8 In contrast to other cross* To whom correspondence should be addressed. Tel: +1+613-9903644. E-mail: [email protected]. † National Research Council Canada. ‡ Current address: CooperVision, Pleasanton, CA 94588-4520. § University of Ottawa, Eye Institute. | University of Tennessee Health Center. ⊥ University of Aberdeen. # University of Ottawa, Department of Ophthalmology.

linkers such as glutaraldehyde and diisocyanates, WSCs do not themselves become incorporated as part of the final cross-links in these hydrogels. Hence, there is no possibility of toxic substance release into tissues from subsequent cross-link break down.2 In addition, all unreacted reagents and byproducts from the WSC reaction are water-soluble and thus can be removed easily after gel formation. Although many WSC cross-linked collagen materials have been previously reported, most of them were formed from very dilute collagen solutions (lower than 1 w/v %), resulting in either very soft gels or insoluble collagen suspensions giving very opaque sponges or highly fibrous materials. Optically clear matrices can be beneficial for cell biology studies of the in vitro and in vivo behavior of cell lines, greatly simplifying characterization by microscopy, even in thick samples such as ocular implants. In recent work12 on corneal performance of a WSC cross-linked type I porcine collagen implant (9 w/w % collagen), we have shown that in vitro (using established cell lines) and in vivo (rabbit and pig lamellar keratoplasty) these sterile matrices have excellent biocompatibility. Briefly, our porcine collagen matrices allowed epithelial overgrowth, followed by normal stratification in vitro. Furthermore, porcine implants recruited in vivo epithelial and stromal cells from each host animal, leading to integration with the host tissue. In pigs, these implants promoted growth of functional neurites as shown by restoration of touch sensitivity. Porcine implants in pigs and rabbits (24 animals in total) retained their original thicknesses and maintained high optical clarity for over 6 months for these ongoing, in vivo tests as shown by slit lamp, confocal microscopy, and topography measurements. Here, we report the preparation of these clear collagen matrices and some of the properties critical for cell overgrowth and infiltration. To our knowledge, this work is the first reported use of WSC chemistry to cross-link collagen solutions at high concentrations

10.1021/bm060160o CCC: $33.50 © 2006 American Chemical Society Published on Web 05/10/2006

1820

Biomacromolecules, Vol. 7, No. 6, 2006

(5-20 w/w %), including recombinant human collagen (rhc), to produce transparent and robust hydrogels suitable for in vitro and in vivo use in cell biology and TE.

Materials and Methods Materials. The WSC, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), N-hydroxysuccinimide (NHS), and albumin fluorescein isothiocyanate conjugate (FTTC-albumin) were purchased from SigmaAldrich (Oakville, Ontario). Acid freeze-dried, type I porcine atelocollagen powder was from Nippon Ham or Koken (both in Tokyo, Japan) and type I rhc in acid solution (0.3 w/w %) was from FibroGen, Inc. (South San Francisco, CA), produced by transfected yeast cells (a strain of Pichia pastoris).5,8 Koken specify that their collagen is 95% collagen I and 5% collagen III. PBS (phosphate buffer saline, pH ) 7.2) was prepared from the dry powder (Invitrogen Canada, Inc., Burlington, Ontario). MilliQ deionized water (Millipore, Billerica, MD) was used throughout. All other reagents and solvents were of analytical grade and used as received. Both porcine collagens gave very similar chemical and physical performance and are not differentiated here. Methods. Collagen Hydrogel Preparation. The acid freeze-dried collagen powders were dissolved in water to give 5-20 w/w % concentrations. Solution required stirring for 24 h or longer at 4 °C with a powerful magnetic stirrer, especially for 20 w/w % solutions. Each resulting acidic collagen solution was loaded into a plastic syringe and suspended air bubbles removed by centrifugation at 4 °C to give clear, bubble-free viscous liquids, ready for dispensing and for mixing with the cross-linking reagents. Only acid freeze-dried collagen powders were found to dissolve completely at 4 °C at up to 20 w/w % to give clear solutions. Neutral freeze-dried collagen samples were all found to give opalescent solutions, possibly because of spontaneous reaction of collagen’s free amine groups with its acid groups during freezedrying. Under acidic conditions, these amine groups are protonated, preventing the cross-linking reaction. The very dilute, acidic rhc solution was concentrated to 10 w/w % by vacuum evaporation at 4 °C, then loaded into syringes as for the porcine collagen samples. To cross-link collagen, an aliquot of a collagen solution (5-20 w/w %, 0.5-1.0 mL) was loaded into a syringe mixing system free of air bubbles.4,9 The pH of the collagen solution was adjusted to 5 ( 0.5 (optimum for EDC chemistry) by injection of microliter quantities of 1.0 M aqueous NaOH, followed by thorough mixing under ice water to reduce heating resulting from the energy of mixing. Calculated volumes of aqueous EDC and NHS solutions (both at 10 w/v %, EDC/ NHS ratio 2:1 in all reactions) were added from a second syringe through a septum in the syringe mixing system and mixed thoroughly with the collagen solution, then cooled in an ice/water bath. This homogeneous solution was immediately dispensed into polypropylene lens molds (12 mm diameter; 80, 350, 440, or 500 µm spacing) and cured at 100% humidity (at 21 °C for 24 h and then at 37 °C for 24 h). Most cross-linked, cornea-shaped hydrogel samples were removed from the molds after soaking in PBS (pH ) 7.2) for 2 h or longer before prying the molds apart. The gels were washed for 4 h in 3 batches of PBS at 4 °C. These fully hydrated hydrogels were then stored in chloroform-saturated PBS to maintain sterility. Storage of hydrogels in chloroform-saturated PBS had no detectable effect on clarity or tensile properties, even after 12 month storage. For comparison purposes, some collagen gels were prepared without addition of a chemical cross-linker, relying on collagen’s spontaneous self-gelation propensity (fibrilogenesis) upon incubation of solutions at 37 °C under close to neutral pH conditions.9 The EDC-to-collagen stoichiometry is reported in this paper as an [EDC/-NH2] factor, which is the ratio of gram equivalents of EDC to gram equivalents of collagen’s primary, -amine groups (from lysine and hydroxylysine residues) in each gel formulation. The gram equivalents of free amine groups in a collagen sample were calculated

Liu et al. by assuming that on average 34-35 amine groups are present per 1000 amino acid residues, both theoretically13 and from -NH2 measurements (see below). Mechanical Testing. Tensile properties of the hydrogels were measured on an Instron Tensile Testing Machine (model 1123, crosshead speed 10 mm/min) by suture pull out (Ethilon, nylon 10/0 sutures, 28-µm-diameter black monofilaments from Johnson and Johnson, Markham, Canada) as described previously.9 This tensile test avoided jaw breaks produced in the fully hydrated hydrogels during direct clamping and drawing. Tensile measurements were performed in triplicate on each gel formulation. Although the failure mode in suture pull out testing is a complex mix of tensile and tearing failure, this method simulates conditions often found during surgical implantation of corneal matrices. Amine Reaction Quantification. The amine content in gels was measured by the trinitrobenzenesulfonic acid (TNBS) method.14 Briefly, each collagen hydrogel cross-linked by EDC/NHS at different [EDC/NH2] levels was put into a separate vial with 4% NaHCO3 and 0.5% TNBS and heated for 4 h at 40 °C, followed by hydrolytic breakdown of the collagen with 6 M HCl at 60 °C. The absorbance of the clear hydrolysate at 345 nm was measured (Hewlett-Packard 8453 spectrophotometer) against a blank, treated in a similar manner but containing no protein. The -NH2 content was calculated from the literature molar absorptivity of 1.46 × 104 cm1 mole-1 for TNBS-lysine.14 Each reacted amine percentage was obtained from

100 × (1 - [-NH2 content post-cross-linking]/ [-NH2 content pre-cross-linking]) Hydrogel Hydration LeVels. Some hydrogel implants were removed carefully from their molds without presoaking in PBS, immediately weighed, and then placed in PBS to equilibrate to constant weight. Post-gelation increase in water uptake (%) is defined as

[(gel wt at equilibrium hydration) (gel wt direct from mold)] × 100/(gel wt direct from mould) The overall (or equilibrium) degree of hydration of the hydrogel implants is defined as that measured after demolding under PBS followed by gel equilibration in PBS for at least 24 h at 4 °C and 2 h at room temperature prior to testing. Hydrogels were removed from the PBS, their surfaces gently tamped dry and then immediately weighed. These samples were then dried at room temperature under vacuum for 24 h or longer to constant weight. The degree of overall hydration was calculated from

(wet weight - dry weight)/(dry weight) To measure the rehydration ratio of fully dried samples, weighed fully hydrated implants were vacuum-dried at room temperature to constant weight, followed by re-equilibration in PBS. The rehydration ratio (%) is defined as

(wt of gel after rehydration) × 100/(wt of gel at equilibrium hydration) Thermal Analysis. Collagen solutions and hydrogels were examined by differential scanning calorimetry (DSC, TA Instruments DSC 2920) after calibration with indium standards. Each preweighed sample of solution or fully hydrated hydrogel (about 15 mg) was placed in an aluminum DSC sample pan, and then, the pan was hermetically sealed. Sealed pans were heated at 2 °C/min from 20 to 80 °C. The denaturing temperature (Td) at the maximum of the endothermic peak and enthalpy of denaturing (∆Hd) were measured. Optical Properties. As described previously,9 transmission and backscattering measurements were made in vitro at 21 °C both for white light (quartz-halogen lamp source) and for narrow spectral regions

Porcine and Recombinant Human Collagen Matrices (centered at 450, 500, 550, 600, and 650 nm) for hydrogels and rabbit or human corneas. The latter were from the Eye Bank of Canada, Toronto, and all five were healthy corneas (center thickness 550 ( 5 µm) from Caucasians, stored for e7 days at 4 °C. Because of the dependence of scatter intensity and asymmetry upon the size of the scattering centers, absolute calibration is extremely difficult, and our reported data can only be used for intersample comparison. Refractive indexes of solutions and hydrogels (fully hydrated in PBS) were measured on samples placed between the prisms of an Abbe´ refractometer (Bellingham and Stanley, U.K.) at room temperature, using sodium D-line illumination. Glucose and Albumin Permeability. Diffusion permeability studies were carried out at 35 °C (the human cornea’s normal, physiological temperature) using two-compartment diffusion chambers with either air flow or mechanical stirring as described previously.12 Briefly, each hydrogel (440 µm in thickness) was used as the membrane between the permeate chamber and the PBS-filled (receptor) chamber. The permeate chamber was filled with either a glucose solution or a bovine, fluorescein isothiocyanate (FITC)-labeled albumin solution. The receptor chamber was sampled at 15 min intervals followed by spectrophotometric analysis using either a glucose assay kit15 or a fluorophotometric analysis procedure for FITC-labeled albumin (495 nm excitation, 519 nm emission).12 Hydraulic Permeability. Flow of PBS through each hydrogel under pressure was measured as reported previously.9 Average pore sizes were derived from the hydraulic permeation coefficient (Ks) and the specific water content of each gel as reported previously for neurite extension studies.16 Statistics. All samples were measured for N g 3. Errors are expressed as ( standard deviations. Biocompatibility. Some in vitro and in vivo procedures have been described previously.12 In brief, for in vitro testing, immortalized human corneal epithelial cells were seeded at ∼8800 cells/ cm2 on top of hydrogel pieces (N ) 4, each 0.57 cm2). Time to complete colonization of the hydrogel surface by epithelial cells in KSFM and then SHEM media was evaluated. Neurite growth from dorsal root ganglia was visualized on and in gels by immunofluorescence. In vitro biodegradation of hydrogels and fresh human cornea material was followed gravimetrically, using the approach of Angele at al.7 Collagen hydrogels cross-linked with EDC and human cornea samples were immersed in DMEM solution augmented to 5 mM in Ca2+ ions for 15 h, then each sample transferred to a fresh DMEM/ Ca2+ solution containing 5.0 units/mL of collagenase Type I (from clostridium histolyticum, Gibco-Invitrogen, Burlington, Ontario). During incubation at 37 °C, samples (n ) 4 for each condition) were sequentially removed, treated in EDTA to block enzyme activity, extracted by ×3 in water to remove salts and nutrients, and then vacuum-dried to constant weight. Biodegradation was followed gravimetrically up to ∼50% decrease in sample dry weight, at which point all samples were too weak for further handling. To examine direct suturability of our implants under surgical conditions, collagen matrices were implanted by penetrating keratoplasty (PK) on live mice (N ) 3). On each mouse, a 1.5-mm-diameter, fullthickness cornea center was removed (∼100 µm in thickness) and replaced with a 2-mm-diameter implant (80 µm thickness). The implant was secured with one continuous running suture (11-0 Ethilon, black monofilament, polyamide suture) which pierced both the implant and the adjacent corneal rim. Sutures were left in the transplanted eye for the whole observation period (120 days).

Results and Discussion In comparison to bovine collagen, porcine collagen implants are expected to be less prone to causing health problems (allergic responses, prion-transmitted diseases, etc.). Furthermore, the use of rhc implants will reduce immunological or allergic reactions and completely prevent the transmission of animal-related

Biomacromolecules, Vol. 7, No. 6, 2006 1821

infectious agents (especially viral and prion-based).8 Because of the expected biodegradation in vivo of natural collagen I following surgery, cross-linking is necessary to allow a useful implant lifetime or to allow time for tissue regeneration by recruitment of cells and neurite extension from the surrounding host tissue. WSCs are very effective collagen cross-linkers. This cross-linking procedure involves the activation of the carboxylic acid groups in collagen by EDC to give an O-acylisourea derivative, but to minimize side reactions, N-hydroxysuccinimide (NHS) is often used as a coreactant.1 NHS converts the initial O-acylisourea reaction product into an activated oxysuccinimide ester. This ester intermediate in turn reacts smoothly and rapidly with primary amine groups in proteins at pH ≈ 5 in aqueous solutions to form the final, stable amide bond, with elimination of N-hydroxysuccinimide.18 In atelocollagen I, the concentrations of carboxylic acid and -NH2 groups are 75 and 35 groups/ 1000 amino acid repeats, respectively, assuming complete cleavage of pro- and telopeptide end groups during collagen purification.13 Chemical identification of EDC-generated, amide cross-links in collagen is very difficult, because only a small number of these new links may be produced in a matrix already highly populated with amide links of the biopolymer’s backbone. However, cross-linking by EDC can be indirectly indicated by gelation and effects on the degree of hydration, by the increase in tensile strength and aggregate stiffness, by biodegradation resistance, by denaturing resistance from DSC, and by consumption of collagen’s pendant amine (-NH2) groups. The natural cornea is a collagen I hydrogel (constituting ∼15% of the fully hydrated weight) cross-linked through glucosaminoglycans (GAGs), which also help control the hydration level of the gel.19 Vision deterioration or loss from corneal damage afflicts many millions of people worldwide and is currently treated by lamellar or penetrating keratoplasty allografts of fresh, human donor corneal tissue.20 However, there is a world shortage of acceptable donor corneas, exacerbated by the storage life of the viable living tissue being only 1 to 4 weeks depending on the storage procedure.21 Consequently, there is an urgent need to develop practical, long-term-storable alternatives to the use of human donor tissue, especially when the risk of disease transmission is taken into account. Furthermore, rejection of donor corneal tissue is quite high, reaching 64% at 5 years after implantation, and innervation is very slow.22,23 The cornea is the main refractive element of the eye, and as such has several key properties and criteria essential for any artificial replacement. These criteria include high optical clarity (high light transmission and low scatter) and appropriate refractive power (controlled curvature and refractive index), adequate biocompatibility, and toughness to withstand surgical procedures and in-use wear and tear.9,10 Although this article is focused on the characterization of the gels, some brief examples of performance in vitro and in vivo suturability are included in addition to the detail already published.12 Biocompatibility. Biocompatibility is a critical area for any material for TE. It encompasses a complex array of interacting parameters, including cell compatibility, biodegradation resistance, nontoxicity, nonimmunogenicity, and noninflammatory properties. Selection of a TE material is always an optimization of these parameters, together with mechanical properties. Figure 1A illustrates in vitro, surface colonization by epithelial cells (confluence, that is, complete colonization of the whole surface, occurred in 4-5 days). Epithelial stratification is not shown here, but was previously confirmed in vivo by histology on cross-sections.12 Figure 1B shows neurite extension over a gel,

1822

Biomacromolecules, Vol. 7, No. 6, 2006

Liu et al.

Figure 2. Effect of the collagen solution concentration on tensile properties of hydrogels, [EDC/-NH2] ) 0.5. (A) Load at rupture. (B) Apparent stiffness. (C) Elongation at rupture.

Figure 1. Biocompatibility of collagen-EDC polymers. (A) In vitro corneal epithelial cell attachment and growth to confluence over 5 days (bar ) 100 µm) on 9% collagen gel, [EDC/-NH2] ) 0.5 equiv/ equiv. (B) In vitro nerve outgrowth from a dorsal root ganglion over the hydrogel surface (bar ) 100 µm) on 9% collagen gel, [EDC/NH2] ) 0.5 equiv/equiv. (C) Mouse cornea immediately after PK operation and surgical suturing of a hydrogel implant (9 w/w % collagen, [EDC/-NH2] ) 0.5 equiv/equiv, 2 mm diameter, 80 µm thickness). Continuous running 11-0 suture, 10 bites. Bar ) 1 mm.

and Figure 1C indicates successful surgical suturing of a 80 µm hydrogel implant during PK on a representative mouse cornea. We have previously reported12 successful in vivo, deep, lamellar keratoplasty on miniature pigs and on rabbits by suturing an implant in place with a Zirm retention bridge overlay of sutures.17 However, in this procedure, sutures overlayed the implant but pierced only the surrounding cornea. This prior in vivo work12 on pigs and rabbits provided the rational for the present detailed study to evaluate the underlying properties which lead to the novel in vivo performance. One formulation (9 w/w % collagen and an [EDC/-NH2] value of 0.5 equiv/ equiv) was selected for all in vivo studies on pigs, rabbits, and mice based on the strength and adequate elongation of the resulting implant matrix (Figures 2 and 3). Our previously reported12 in vivo and cell biology performance of our collagen-EDC matrices showed that new nerve

neurites grew to traverse the implant and be functional as confirmed by restoration of touch sensitivity of the eye. In vivo slit lamp and microscopy images recorded from 2 to 6 months postimplantation in 24 animals (16 rabbits and 8 mini-pigs) showed absence of inflamation without the use of any postoperative drugs (apart from an initial pain medication). The implants were optically clear by slit lamp. There were no differences between the implanted and nonoperated eyes in each animal by slit lamp biomicroscopy, and there was no change in refractive power, overall corneal thickness, or topography.12 Biodegradation resistance is important for long-term performance. Optically clear materials with a very high resistance to biodegradation, such as pHEMA and polyperfluoroethers, have been promoted as beneficial for corneal implants, but these implants did not promote normal epithelial overgrowth and stratification.24-26 More cell-friendly materials such as collagen do allow re-epithelialsation and, for our materials, nerve regeneration (Figure 1A,B), but are expected to biodegrade in the long term, even if cross-linked. In vitro biodegradation rates (constant up to ∼50% loss in dry weight) by collagenase are listed in Table 4 for several 9 w/w % collagen implants as a function of cross-linking stoichiometry ([EDC/-NH2] value). (Biodegradation rates for matrices prepared under the crosslinking chemistry conditions but in the absence of EDC/NHS could not be measured because of the very weak nature of these highly opaque materials.) Although high stoichiometries lead to slower biodegradation as expected, such samples were found to be too brittle for surgical implantation. At [EDC/-NH2] ) 0.5, the biodegradation rate is about 10 times faster than the native cornea under the same test conditions. However, what this rate difference will mean in practice for a corneal implant is currently unknown. Extrapolation from the in vitro test conditions to in vivo conditions is currently impossible, and

Porcine and Recombinant Human Collagen Matrices

Figure 3. Effect [EDC/-NH2] levels on tensile properties of hydrogels. Collagen concentration in gels ) 9 w/w %. (A) Load at rupture. (B) Apparent stiffness. (C) Elongation at rupture. * rhc.

the true test of long-term performance can only come from an in vivo study. Our recent 6 month in vivo study12 clearly indicated retention of overall corneal thickness, refractive power, and surface topography, very similar to the unoperated (contralateral) eye of each animal, with negligible haze development. Whether this performance results from absence of significant biodegradation over this time period or from progressive implant biodegradation with concomitant regeneration of collagen by the recruited stromal cells remains to be established and is under investigation. Mechanical Properties. The influence of cross-linking on the mechanical properties of fully hydrated hydrogels was determined with EDC/NHS molar ratio of 2:1 and a pH of 5.0 to 5.5 for all reactions. For [EDC/-NH2] fixed at 0.5, the load at break of collagen hydrogels increased from ∼4 g-force to ∼28 g-force when the initial collagen solution concentration increased from 5 to 20 w/w % (Figure 2A) and the apparent stiffness increased from 2.3 (g-force/mm) to 5.5 (g-force/mm) for this series (Figure 2B), but when the collagen concentration was increased from 15 to 20 w/w %, the apparent stiffness change was very small. The elongation at break of the hydrogel increased from 2.0 mm to 5.5 mm when collagen concentration increased from 5 to 20 w/w % (Figure 2C). At a fixed collagen concentration in gels (9 w/w %), increasing [EDC/-NH2] resulted in only a slight change in hydrogel strength up to [EDC/-NH2] ) 0.5, followed by a decrease (Figure 3A). The apparent stiffness increased progressively with increasing [EDC/-NH2] up to [EDC/-NH2] ) 1.0 (Figure 3B). The elongation at break of the hydrogel decreased steadily from 5.7 to 1.7 mm as [EDC/-NH2] increased from 0.125 to 1.0 (Figure 3C). These results suggested that the gels made at [EDC/-NH2] values greater than 1.0 will be brittle (greater stiffness, but lower elongation and load at break).

Biomacromolecules, Vol. 7, No. 6, 2006 1823

The increase in tensile strength and apparent stiffness and decrease in elongation at break caused by cross-linking collagen can obviously be explained by the introduction of the amide intermolecular bridges in the matrix. These bridges will hinder the collagen helices, microfibrils, and fibers from orienting during the draw process. Porcine collagen and rhc showed quite similar tensile behavior at comparable collagen concentrations and EDC stoichiometries (Figure 3). Although just strong enough to be handled with forceps, the collagen thermogel (without EDC/NHS) was too weak to allow measurement of tensile properties. Amine Group Reaction. The TNBS method is specific for the primary amine group of lysine and hydroxylysine.27 Our measured value for the -NH2 content of the porcine atelocollagen (34 ( 3 equiv/1000 amino acid repeats of collagen) is consistent with the theoretical and experimental values for mammalian collagen.1,7,13 Measurement of -NH2 content in collagen before and after EDC/NHS cross-linking showed that the -NH2 had been reduced, consistent with cross-link formation in the hydrogel, although only a relatively small percentage of the -NH2 groups were consumed. The percentage of reacted -NH2 groups in gels with 9 w/w % collagen increased from 6% to 24% when [EDC/-NH2] increased from 0.25 to 1.0 (Figure 4). However, when [EDC/-NH2] was further increased to 2.0, only a 21% consumption of the amine groups was detected. This decrease might result from nonuniform mixing of the very viscous collagen solution with EDC as a result of the very fast gelation (